(12a) Accounting for Safety System Activation within Economic Model Predictive Control | AIChE

(12a) Accounting for Safety System Activation within Economic Model Predictive Control


Albalawi, F. - Presenter, University of California, Los Angeles
Wu, Z., University of California Los Angeles
Zhang, Z., University of California, Los Angeles
Durand, H., University of California, Los Angeles
Christofides, P., University of California, Los Angeles
Chemical plants utilize a set of barriers against process accidents consisting of the control system and the safety system (e.g., the alarm, emergency shut-down, and safety relief systems) [1]. Recent advances in chemical process safety utilizing a systems engineering perspective [2] have sought to incorporate safety considerations within economic model predictive control (EMPC) [3] through safety-based constraints (either Lyapunov level set-based constraints [4] or constraints based on a safety metric termed the Safeness Index [5]). These designs have assumed that the process model does not change throughout the prediction horizon, meaning that no actions of the safety system (which are typically on/off-type actions such as bringing a valve from its fully open to fully closed position) are considered to be taken during process operation. However, the activation of the safety system should be accounted for within EMPC design because the actions of the safety system change the process dynamics and potentially the input availability (e.g., in the case that a valve actuated by the safety system is in series with a valve actuated by the control system), and thus the state predictions from the EMPC should account for activation of the safety system to avoid significant plant-model mismatch throughout the prediction horizon that may cause the EMPC to choose less suitable control actions (from both an economics viewpoint and a safety viewpoint) than it would choose if it was aware of the change in the plant due to safety system activation.

Motivated by the above considerations, we develop an EMPC design that accounts for the activation conditions of the safety system and changes the process model when the safety system is predicted to be activated at any time throughout the prediction horizon to reflect the dynamics of the process after safety system activation. We consider modeling of the automated actions of both the emergency shutdown system and the safety relief system within the EMPC. We investigate the impact that the foreknowledge of the safety system activation has on the state and input trajectories of a chemical process under such a control design compared to the case that the EMPC is unaware of the safety system activation. We discuss the impact of the plant-model mismatch that may occur due to disturbances if the state predictions utilized for choosing the control actions indicate activation of the safety system but it does not activate for the process with disturbances (similarly, we discuss the impact of the safety system activating when it was not predicted within the EMPC to do so). We investigate the addition of constraints to the EMPC as the state predictions approach the safety system activation levels to encourage it to choose control actions that will prevent the safety system activation. The closed-loop modeling strategy developed that includes the safety system activation conditions may also provide a systematic methodology for identifying safety system triggers and thus for coordinating the control and safety systems.

[1] Marlin T. Operability in Process Design: Achieving Safe, Profitable, and Robust Process Operations. McMaster University in Ontario, Canada, 2012.

[2] Leveson NG, Stephanopoulos G. A system-theoretic, control-inspired view and approach to process safety. AIChE Journal. 2014;60:2-14.

[3] Ellis M, Durand H, Christofides PD. A tutorial review of economic model predictive control methods. Journal of Process Control. 2014;24:1156-1178.

[4] Albalawi F, Alanqar A, Durand H, Christofides PD. A feedback control framework for safe and economically-optimal operation of nonlinear processes. AIChE Journal. 2016;62:2391-2409.

[5] Albalawi F, Durand H, Christofides PD. Process operational safety using model predictive control based on a process Safeness Index. Computers & Chemical Engineering. in press.